Updated Readme with SDXL images
Browse files
README.md
CHANGED
|
@@ -8,10 +8,10 @@ tags:
|
|
| 8 |
|
| 9 |
# IP Composition Adapter
|
| 10 |
|
| 11 |
-
This adapter for Stable Diffusion 1.5 is designed to inject the general composition of an image into the model while mostly ignoring the style and content. Meaning a portrait of a person waving their left hand will result in an image of a completely different person waving with their left hand.
|
| 12 |
|
| 13 |
-
###
|
| 14 |
-
|
| 15 |
|
| 16 |
### Thanks
|
| 17 |
|
|
@@ -19,13 +19,31 @@ I want to give a special thanks to [POM](https://huggingface.co/peteromallet) wi
|
|
| 19 |
|
| 20 |
## Usage
|
| 21 |
|
| 22 |
-
Use just like other IP+ adapters from [h94/IP-Adapter](https://huggingface.co/h94/IP-Adapter)
|
|
|
|
|
|
|
| 23 |
|
| 24 |
### How is it different from control nets?
|
| 25 |
|
| 26 |
Control nets are more rigid. A control net will spatially align an image to nearly perfectly match the control image. The composition adapter allows the control to be more flexible.
|
| 27 |
|
| 28 |
-
## Examples
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 29 |
|
| 30 |
<img src="https://huggingface.co/ostris/ip-composition-adapter/resolve/main/resources/Screenshot%20from%202024-03-16%2013-06-32.jpg" alt="1" width="100%"/>
|
| 31 |
|
|
|
|
| 8 |
|
| 9 |
# IP Composition Adapter
|
| 10 |
|
| 11 |
+
This adapter for Stable Diffusion 1.5 and SDXL is designed to inject the general composition of an image into the model while mostly ignoring the style and content. Meaning a portrait of a person waving their left hand will result in an image of a completely different person waving with their left hand.
|
| 12 |
|
| 13 |
+
### Follow Me
|
| 14 |
+
I do a lot of experiments and other things. To keep up to date, follow me on [Twitter](https://twitter.com/ostrisai).
|
| 15 |
|
| 16 |
### Thanks
|
| 17 |
|
|
|
|
| 19 |
|
| 20 |
## Usage
|
| 21 |
|
| 22 |
+
Use just like other IP+ adapters from [h94/IP-Adapter](https://huggingface.co/h94/IP-Adapter). For both SD1.5 and SDXL variants, use the CLIP vision encoder ([CLIP-H](https://huggingface.co/h94/IP-Adapter/tree/main/models/image_encoder))
|
| 23 |
+
|
| 24 |
+
You may need to lower the CFG to around 3 for best results, especially on the SDXL variant.
|
| 25 |
|
| 26 |
### How is it different from control nets?
|
| 27 |
|
| 28 |
Control nets are more rigid. A control net will spatially align an image to nearly perfectly match the control image. The composition adapter allows the control to be more flexible.
|
| 29 |
|
| 30 |
+
## SDXL Examples
|
| 31 |
+
|
| 32 |
+
<img src="https://huggingface.co/ostris/ip-composition-adapter/resolve/main/resources/Screenshot%202024-03-19%20195340.png" alt="1" width="100%"/>
|
| 33 |
+
|
| 34 |
+
<img src="https://huggingface.co/ostris/ip-composition-adapter/resolve/main/resources/Screenshot%202024-03-19%20194130.png" alt="1" width="100%"/>
|
| 35 |
+
|
| 36 |
+
<img src="https://huggingface.co/ostris/ip-composition-adapter/resolve/main/resources/Screenshot%202024-03-19%20193948.png" alt="1" width="100%"/>
|
| 37 |
+
|
| 38 |
+
<img src="https://huggingface.co/ostris/ip-composition-adapter/resolve/main/resources/Screenshot%202024-03-19%20193802.png" alt="1" width="100%"/>
|
| 39 |
+
|
| 40 |
+
<img src="https://huggingface.co/ostris/ip-composition-adapter/resolve/main/resources/Screenshot%202024-03-19%20193632.png" alt="1" width="100%"/>
|
| 41 |
+
|
| 42 |
+
<img src="https://huggingface.co/ostris/ip-composition-adapter/resolve/main/resources/Screenshot%202024-03-19%20192659.png" alt="1" width="100%"/>
|
| 43 |
+
|
| 44 |
+
<img src="https://huggingface.co/ostris/ip-composition-adapter/resolve/main/resources/Screenshot%202024-03-19%20192445.png" alt="1" width="100%"/>
|
| 45 |
+
|
| 46 |
+
## SD 1.5 Examples
|
| 47 |
|
| 48 |
<img src="https://huggingface.co/ostris/ip-composition-adapter/resolve/main/resources/Screenshot%20from%202024-03-16%2013-06-32.jpg" alt="1" width="100%"/>
|
| 49 |
|